EulerMaruyama with explicit batch dimension parameter

[Biosins]

Hello all,

I've been learning probabilistic programming for a couple of weeks now and playing around with a couple of toy examples, mainly focusing on approximating or solving SDEs

There is a distribution for the EulerMaruyama method, but it seems like it has issues when providing parameters that already have an batch dimension, for example in the case of solving two coupled SDEs.

Is there an example of correctly shaping the parameters for this case, or is it a limitation of that distribution at the moment?

[ricardoV94]

What do you mean coupled? If they're related those are not batch dimensions but instead a multivariate version of the distribution. If they are just two and independent you could just define two separate EurelMaruyama distributions